DRM_CAP_PRIME(9)

πŸ“¦ version 7.3.1, in nixpkgs unstable

πŸ’… DRM_CAP_PRIME πŸ’…

Okay so bestie, this whole page lives in the Kernel API Manual β€” that's the 9 hanging off the end of the name, iykyk. It's a teeny tiny one, literally a one-capability wonder, but we're gonna give it the full main-character treatment. 🫢


✨ NAME ✨

DRM_CAP_PRIME β€” and that's it, that's the whole name line, no cap. The original leaves the description blank, so we are not inventing one. The name is the moment. πŸ‘‘


πŸ¦‹ SYNOPSIS πŸ¦‹

DRM_CAP_PRIME

Just the bare identifier, sitting there being iconic. No arguments, no flags, no drama. πŸ’…


🌸 DESCRIPTION 🌸

So this thing? It's a bitfield of supported PRIME sharing capabilities. That's the whole vibe. πŸŽ€

If you wanna know what the individual bits mean, go look at DRM_PRIME_CAP_IMPORT and DRM_PRIME_CAP_EXPORT β€” they're the actual receipts, not this page.

Now, starting from kernel version 6.6, plot twist: both DRM_PRIME_CAP_IMPORT and DRM_PRIME_CAP_EXPORT are always advertised. Like, every single time. No more vibe check, the answer is always yes. ✨

Also PRIME buffers are exposed as dma-buf file descriptors β€” yes, the fancy shared-memory-ish fd thing. See :ref:prime_buffer_sharing`` for the deep lore. 🧚


πŸ”₯ SEE ALSO πŸ”₯

Okay this list is long, but every single one of these deserves their moment, so here we go, no skipping, we don't do that here: πŸ’–


Lowkey the most wholesome part: this page traces back to ./tools/perf/trace/beauty/incl... (it gets cut off in the footer, bestie, we don't make stuff up here πŸ™ˆ) and it's dated October 2026. Version vibes: DRM_CAP_PRIME(9). Slay, bye! πŸ’œπŸŽ€πŸ¦‹
πŸ“¦ where this page comes from